diff --git a/Grid/tensors/Tensor_SIMT.h b/Grid/tensors/Tensor_SIMT.h
index 672f385f..0a7d3382 100644
--- a/Grid/tensors/Tensor_SIMT.h
+++ b/Grid/tensors/Tensor_SIMT.h
@@ -65,7 +65,8 @@ void coalescedWriteNonTemporal(vobj & __restrict__ vec,const vobj & __restrict__
 #else
 
 
-#ifndef GRID_SYCL
+//#ifndef GRID_SYCL
+#if 1
 // Use the scalar as our own complex on GPU ... thrust::complex or std::complex
 template<class vsimd,IfSimd<vsimd> = 0> accelerator_inline
 typename vsimd::scalar_type